How AI can help reduce food waste
Check out all the on-demand sessions from the Intelligent Security Summit here.
Computers have a bad rap when it comes to saving the planet. Cryptocurrencies, due to the highly inefficient tech involved, have been consuming as much electricity as the entire country of Sweden. Elon Musk has repeatedly warned about a Terminator-style apocalypse likely to be brought on by artificial intelligence (AI). And yet, like any tool, AI has a tremendous potential to be good for the planet — and this future isn’t as far as it seems.
Today, let’s examine one aspect of this potential: Reducing carbon emissions from food-related systems. According to Nature, these account for a third of total emissions; the growing world population points to the increasing significance of this factor with time.
Computers are wonderful at keeping track of a myriad of factors and adjusting outputs without human intervention. There are at least two food system-related tasks for which this is highly applicable: Reducing food waste and driving consumption of foods that are better for the environment. Let’s examine each in detail.
Reducing food waste at home
According to the USDA, 21% of food that consumers bring to their homes ends up wasted, and another 10% is thrown out at the grocery store/warehouse. Let’s look at the root causes of this waste.
Intelligent Security Summit On-Demand
Learn the critical role of AI & ML in cybersecurity and industry specific case studies. Watch on-demand sessions today.
A significant factor is consumers not knowing what to do with the food that caught their eye at the store. Maybe it was on sale; maybe a portion of the item was used for a recipe, and the leftovers don’t offer a good path forward. Whenever there’s no plan — no recipe to make with a particular grocery item — the chance of it going to waste goes up. This is particularly true for items with short shelf lives, such as vegetables and proteins.
But what if the grocery shopping paradigm was shifted from focusing on individual grocery items to focusing on recipes? Each item in the fridge would then have a “plan” around it; as long as the recipes are what the family wants, the items will all get eaten.
This paradigm, combined with AI that zooms into family food preferences and recommends recipes each family would enjoy, has been quite powerful. Recipe-based shopping is something Instacart and Amazon are embracing as well; there’s no reason physical grocery stores can’t too.
Moreover, instead of thinking of recipes as standalone, grocery retailers should consider how consumers can reuse ingredients across recipes for the week. For example, if one recipe in the customer’s cart calls for parsley as a garnish, a complementary salad recipe can use the rest of the parsley bunch. This saves customers money and reduces the chance that unused parsley goes to waste.
This task — pairing up complementary recipes to make the best use of leftovers — is perfect for AI.
Reducing food waste at the grocery store
Much waste in the grocery store and warehouse is the result of overstocking. Despite supply chain systems being fully computerized, and marketplace incentives to improve, the USDA still pins retail losses at 10%. Consumer behavior is quite difficult to predict as long as the business is based on consumers browsing virtual or physical aisles and picking the grocery items they want.
What if this model was reversed? What if consumers didn’t directly pick the grocery items or even recipes that they want; what if they stated their broad food preferences and an agent acting on their behalf (a human or an AI) did the shopping for them? Provided this agent does a good job representing the consumer’s needs, the agent can also be made aware of the inventory levels at the retailer; they could then make substitutions that have no impact on consumer satisfaction but prevent spoilage.
Besides the obvious benefit for the planet, reduction in waste creates a more profitable business and allows some savings to be passed on to the consumer. When typical grocery store margins are in single digits, these savings add up — especially in an inflationary environment.
Eating what’s better for the environment
Following a diet that has a low carbon footprint is a surprisingly counterintuitive task for humans. According to Our World in Data, local food is often no better than food shipped from continents away. Organic food often has a higher greenhouse gas footprint. Even reducing packaging isn’t the right factor to pay attention to: It’s a tiny fraction of a food’s environmental impact and often lengthens its shelf-life, reducing waste.
It’s too hard to keep track of the latest understanding of what’s actually good and bad, and the research is rapidly evolving. So the cognitive load of keeping up is too high, even for those consumers who care deeply about eating sustainably.
Wouldn’t it be great if there was an autopilot? Something like the ESG investment funds that do the work for you, but in the food realm? Something that would help you do the right thing and send you a quarterly report about how much better you did than the average Joe?
Unlike investing, where you can be as hands-off as you want, this doesn’t work as easily with food. Besides caring about your food’s sustainability properties, you very much care about the taste, allergens, macronutrient content, portion size and a bunch of other factors. Unless you’re vegan, there are plenty of vegan meal options you wouldn’t enjoy, and many vegan, vegetarian and low-carbon omnivore options that you would like.
Understanding all of the customer needs and adjusting recommendations based on a feedback loop (using structured, explicit feedback) is a key enabler here.
Imagine a world where an autopilot for healthy and sustainable eating exists. If this autopilot knows each consumer well, it can confidently nudge some of them towards more sustainable foods — swapping out a beef-based recipe for one with chicken or introducing a vegetables-forward meal to someone who normally tilts heavily towards meat. AI plays a central role in these nudges because each customer’s preference is unique; and because collecting feedback at scale and adjusting recommendations based on it is key to fulfilling all of the objectives.
This concept of micro-nudges is highly relevant. Featuring sustainable options in the shopping experience, along with social proof, can help traditional “browse-the-aisles” retailers help consumers make the right choices. For digital retailers, knowing more about each customer can help optimize relevance against sustainability. In the optimal case, these two variables don’t need to compete.
AI as a force for good
As we’ve seen here, AI-based systems can help reduce greenhouse gas emissions in two profound ways: By reducing food waste and by nudging consumers to eat more sustainable foods. Each of these factors can have a profound impact on the planet in the next decade.
Alex Weinstein is CDO at Hungryroot.
Welcome to the VentureBeat community!
DataDecisionMakers is where experts, including the technical people doing data work, can share data-related insights and innovation.
If you want to read about cutting-edge ideas and up-to-date information, best practices, and the future of data and data tech, join us at DataDecisionMakers.
You might even consider contributing an article of your own!